Anal. Chem.
2000, 72, 3563-3567
Quantification of Polycyclic Aromatic
Hydrocarbons in the NIST Standard Reference
Material (SRM1649A) Urban Dust Using Thermal
Desorption GC/MS
Daniel Waterman, Brian Horsfield,† Franz Leistner,† Keith Hall,‡ and Steve Smith*
Division of Life Sciences, King’s College London, Franklin-Wilkins Building, 150 Stamford Street, London, SE1 8WA, U.K.
A thermal desorption GC/MS technique has been devel-
oped for the quantification of polycyclic aromatic hydro-
carbons (PAHs) in airborne particulate matter using the
NIST Standard Reference Material (SRM1649a) Urban
Dust. The technique was developed using standard lin-
earity tests in order to establish optimum sample weights
and optimum desorption and chromatographic param-
eters. This direct analysis technique eliminates the use
of solvents in the sample preparation (reducing volatile
component losses) and also significantly reduces the
sample preparation time (no extraction procedure). The
technique has been shown to give linearity in terms of the
overall TIC response as well as for a prominent series of
n-alkanes (C20-C33) and 10 NIST priority PAHs, 8 of
which have been quantified. The technique is reported to
be uniquely sensitive (PAH concentrations 2-6 mg kg-1)
and reproducible (MW ) 178-228 SD e0.228 mg kg-1,
e7%; MW ) 252 SD e0.922 mg kg-1, e33%) over the
range of sample weights (1-5 mg). Such sample weights
illustrate that the technique can be equally applied to the
analysis of airborne particulate samples collected over
short time periods (24-48 h) using only commonly used
low-volume collection devices.
The adverse health effects of airborne particulates are well
recognized, and correlation between particulate concentration and
incidents of respiratory problems, in particular, have been well
illustrated in epidemiological studies, e.g., refs 1 and 2. Such
studies have raised concerns regarding the concentration and
nature of airborne particulate pollution and although inroads have
been made in understanding the chemical, biological, and physical
properties of airborne particulates, such work has also succeeded
in defining the problem as an extremely complex one.
* To whom correspondence should be addressed: (e-mail) steve.smith@
kcl.ac.uk; (fax) 44 (0)20 7848 4500.
† Current address: Institute of Petroleum and Organic Geochemistry, For-
schungszentrum Jülich GmbH, D-52425 Jülich, Germany .
‡ Current adress: Hall Analytical Laboratories Ltd., Unit A, Millbrook Business
Centre, Floats Rd., Manchester M23 9YJ, U.K.
(1) Dockery, D. W.; Pope, G. A.; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.;
Ferris, B. G.; Speizer, F. E. N. Engl. J. Med. 1993, 329, 1753-1759.
(2) Pope, C. A.; Thun, M. J.; Namboodiri, M. M.; Dockery, D. W.; Evans, J. S.;
Speizer, F. E.; Heath, C. W. J. Am. Respir. Crit. Care Med. 1995, 151, 669-
674.
10.1021/ac991372x CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/29/2000
Airborne particulate matter consists of particles of variable size
and composition and is often referred to as PM10 (particles of
mean aerodynamic diameter of 10 µm). A greater understanding
of how particles are deposited within the respiratory system3 has
meant interest, particularly in understanding the toxicity mech-
anism, has moved toward smaller fractions of PM2.5, PM1, and
ultrafine fractions (<0.1 µm).
Chemical characterization is important in adding to the
understanding of the toxicity mechanism as well as being vital in
assessing the sources of airborne particulate matter (source
apportionment) and, therefore, is key to implementation of
environmental policies for the reduction of major sources of
particulates. As interest moves toward smaller particles, there is
an increasing need for highly sensitive chemical characterization
techniques that are capable of working with extremely small
samples of particulate matter composed of complex mixtures of
chemical species in low concentrations.
Multiple and variable source factors mean that particulate
composition is extremely complex and variable.4 Methods for the
analysis of the inorganic fraction of particles (elemental analysis,
metals analysis, soluble ions analysis) are well established, e.g.,
ref 5 whereas organic analysis has relied mainly on solvent
extraction techniques (Soxhlet), followed by GC/HPLC analysis.6
When coupled to organic geochemical techniques such as bio-
marker analysis and source/emission profiling, the technique has
proved a useful tool for source apportionment work (for a recent
review, see ref 7).
Although the Soxhlet extraction has been the primary tool of
organic analysis, it can be considered to suffer from a number of
limitations. The efficiency of a Soxhlet extraction depends on the
solubility of the individual organic components in the selected
solvent and the ability of the solvent to penetrate the matrix.
Extraction is often maximized by the utilization of lengthy multiple-
stage extractions, multiple solvent extractions, or physical agitation
such as refluxation or ultrasonication. Furthermore, the relatively
(3) Task Group of Lung Dynamics (Committee IIsICRP). Health Phys. 1996,
12, 173-207.
(4) Source Apportionment of Airborne Particulate Matter in the United Kingdom
(APEG 1999). Report of the Airborne Particles Expert Group on behalf of
the DETR, UK, 1999.
(5) Harrison, R. M.; Jones, M. Sci. Total Environ. 1995, 168, 195-214.
(6) Institute of Petroleum. Determination of soluble organic fraction (SOF) of
diesel particulates on filterssSoxhlet extraction method. IP PM CA/96.
(7) Simoneit, B. R. T. Environ. Sci. Pollut. Res. 1999, 1 (3), 159-169.
Analytical Chemistry, Vol. 72, No. 15, August 1, 2000 3563
large weights of deposit required (typically 8-10 mg) often
necessitates the bulking of samples or high-volume sampling
equipment. Soxhlet extraction also suffers from being inappropri-
ate for low molecular weight, volatile compounds, because they
are either lost in the solvent reduction stage of sample preparation
or are masked by the solvent itself during subsequent gas
chromatographic analysis.
Improved extraction techniques have more recently been
developed such as microwave digestion8 and supercritical fluid
extraction9 that have been shown to provide comparable or
improved extraction yields as well as being less time consuming.
However, both techniques still require a separate extraction stage
prior to analysis and relatively large weights of deposit.
Thermal desorption eliminates the long extraction stage, has
been shown to give reproducible organic fingerprints (including
more volatile compounds), and requires considerably smaller
samples. It was with these factors in mind that we evaluated
thermal desorption as a technique for quantifiable organic analysis
of airborne particulates.
Thermal desorption has already proved to be a valuable tool
in the earth science and energy sectors when coupled with on-
line gas chromatography analysis.10,11 Environmental applications
of the technique have tended to involve absorption of organics
onto solid supports such as Tenax or SPME fibers prior to
desorption and analysis (see ref 12 for a recent review). Typically
contaminants from water13 (studies of wastewater and contami-
nated soils) or volatile organics in air14 have been studied by these
methods.
The technique of direct thermal desorption with on-line GC is
used in the oil industry for analyzing particulates collected directly
from engine exhausts. The technique is not established as a
quantifiable method but is used to examine the unresolved, broad
fuel and lubricant peaks observed using a wide-bore GC column
in order to assess the fuel mix performance.15
Pyrolysis GC/MS, a significantly more energetic technique
involving considerable breakdown and fragmentation of large
molecules and the solid matrix has started to be used in
environmental analysis but has tended to concern studies of
biomarkers in contaminated soils16 and potential applications in
bioremediation work.17,18
EXPERIMENTAL SECTION
Thermal Desorption GC/MS Hardware. The thermal de-
sorption system (illustrated in Figure 1) utilizes a Quantum
MSSV-2 thermal analysis port (GC2 Chromatography and EGO
(8) Wong, M. K.; Gu, W.; Ng, T. L. Anal. Sci. 1997, 13 (55), 97-102.
(9) Oda, J. Bunseki Kagaku 1999, 48 (6), 595-607.
(10) Horsfield, B.; Disko, U.; Leistner, F. Geol. Rundsch. 1989, 78 (1), 361-
374.
(11) Muscio, G.; Horsfield, B.; Welte, D. H. Adv. Org. Geochem.; Øygard, K., Ed.
Org. Geochem. 1994, 22 (3-5), 461-476.
(12) Fox, D. L. Anal. Chem. 1999, 71, 109R-119R.
(13) Grote, C.; Belau, E.; Levsen, K.; Wunsch, G. ACTA Hydrochim. Hydrobiol.
1999, 27 (4), 193-199.
(14) Pankow, J. F.; Luo, W.; Isabelle, L. M.; Bender, D. A.; Baker, R. J. Anal.
Chem. 1998, 70, 5213-5221.
(15) Institute of Petroleum. Analysis of fuel and oil derived hydrocarbons in diesel
particulates on filterssGas Chromatography method. IP PM BZ/96.
(16) Medina-Vera, M. J. Anal. Appl. Pyrolysis 1996, 36, 27-35.
(17) Richnow, H. H.; Seifert, R.; Kästner, M.; Mahro, B.; Horsfield, B.; Tiedgen,
U.; Böhm, S.; Michaelis, W. Chemosphere 1995, 31 (8), 3991-3999.
(18) White, D. M.; Irvine, R. L. Environ. Monit. Assess. 1998, 50 (1), 53-65.
3564 Analytical Chemistry, Vol. 72, No. 15, August 1, 2000
Figure 1. Thermal desorption GC/MS system.
Inc.). The unit comprises a desorption double oven mounted on
the top of a Hewlett-Packard 5890 gas chromatograph, connected
to a short-path interface and a direct heated cryogenic trap filled
with glass beads. The interface has a purge outlet and a nitrogen
back-flush facility for cleaning debris from the system. The control
panel has fully programmable temperature facilities allowing
separate control of the oven, the interface, and the cryogenic trap
as well as programmable temperature ramping and control of the
carrier gas flow rate. The cryogenic trap is cooled using a liquid
nitrogen dewar internal to the GC oven that is lowered on
completion of the desorption. The timing of the trap heating is
integral to the GC temperature program and is monitored by the
control panel.
The desorption oven has an extension to the inlet that allows
ambient temperature purging of samples. The desorption liner is
made of steel and is held in position by an adjustable height arm.
The GC split is positioned after the cryogenic trap giving greater
control of flows, resulting in better reproducibility, prior to
connection to the GC capillary column (Phenomenex ZB-5, 25 m,
0.25 mm i.d., 0.25 µm ft). The GC is interfaced to a Fisons MD800
mass spectrometer run in EI mode and scanned from m/z 40 to
520.
Sample Preparation. The NIST Standard Reference Material
SRM1649a Urban Dust was collected using a baghouse method
near to Washington, DC, and was purchased through the Labora-
tory of the Government Chemist (Teddington, U.K.). The material
is certified for a number of polycyclic aromatic hydrocarbons (mg
kg-1), for a series of PCBs (µg kg-1), and for total organic content
and extractable (Soxhlet) organic content (note: n-alkanes are
not certified).
Samples were prepared by weighing quantities of the urban
dust into a glass-lined stainless steel GC liner (cleaned at 600 °C
for 5 min) containing a small plug (∼1 mg) of glass wool (cleaned
at 350 °C). A second plug of glass wool is positioned on top of
the urban dust, minimizing flow disturbances and consequent
losses or displacement of material from the heated zone of the
desorption oven. Internal standards could then be added as
dichloromethane (DCM) solutions from a microsyringe.
Figure 2. Chromatogram of 2.90 mg of thermally desorbed SRM1649a Urban Dust. Asterisks indicate consecutive n-alkanes from n ) 20 to
n ) 33.

Analytical Procedure. The GC split ratio is set (15:1) with a

column flow rate of 0.8 mL min-1 prior to the liner being placed
in the ambient temperature purge extension of the desorption unit,
and the sample is purged for 1 min. The dewar of the cryogenic
trap is then filled with liquid nitrogen and the trap submerged.
The data acquisition and the GC temperature programs are started
simultaneously, and the liner is lowered into the body of the
desorption oven which is then ramped from 175 to 300 °C at 50
°C min-1. On vaporization, the desorbants are swept by the carrier
gas from the heated zone, minimizing potential thermal decom-
position, and are focused on the trap for the 10-min desorption
period. The trap is then lowered and almost simultaneously flash
heated from -196 to 300 °C in 20 s, flushing the desorbants held
on the trap onto the GC column for analysis. The GC oven is set
at 40 °C for 13 min (including desorption time) and then ramped
at 5 °C min-1 to 300 °C and held for 25 min (total analysis time
90 min).
Carrier gas flows are allowed to equilibrate for ∼5 min between
each stage of the process and confirmed by monitoring flow rates
on a Humonics Veri-Flow 500 precision low-volume flow meter. Figure 3. Chromatographic linearity (R2 ) 0.948) of the Standard
On completion of the GC program, the system is cooled, the liner Reference Material SRM1649a Urban Dust.
is removed, and the inlet system is cleaned of fragments of glass
wool and dust particles by back-flushing with a 2 bar pressure of conditions, versus total sample weight (Figure 3). The plot
nitrogen through the purge outlet valve. All equipment is handled indicates good linearity (R2 ) 0.948) within the range of weights
using cotton gloves, and all implements used in the preparation used, 1-5 mg. As well as the TIC linearity, a series of n-alkanes
are first cleaned with solvent and then in an oven at 350 °C. (C20-C33) labeled in Figure 2, were identified and the chromato-
Linearity was confirmed using different weight samples of the graphic areas plotted against sample weight illustrating good
SRM and plotting integrated areas of the total chromatographic linearity in the microgram range (Figure 4).
area (relative to blank baseline), a series of n-alkanes (C20-C33) Ten NIST priority PAHs were identified (phenanthrene,
and a series of PAHs (MW 178-276), against total sample weight. pyrene, fluoranthene, benz[a]anthracene, chrysene, benzo[b]-
Similarly linearity plots were obtained by desorbing a series of fluoranthene, benzo[e]pyrene, benzo[a]pyrene, benzo[ghi]p-
1-µL volumes of different concentrations of phenanthrene-d10 in erylene, indeno[1,2,3-cd]pyrene) covering molecular weights 178-
DCM injected onto a small quantity of glass wool in a blank liner. 276 and selected from the chromatograms and the integrated areas
Suitable weights (within the linearity) of both Urban Dust and plotted against total sample weight (Figure 5). The plots indicate
phenanthrene-d10 were then utilized for quantification of the PAHs. that within the range of 1-5 mg of urban dust a linear response
for the 10 PAHs is seen. The linearity ranges closely follow that
RESULTS AND DISCUSSION of the total chromatographic linearity range. If then the total urban
Linearity Tests. Chromatograms (example in Figure 2) were dust sample weight (1-5 mg) is converted to PAH weight utilizing
obtained from different weights of samples (1-5 mg) of SRM run the certified values for each PAH, then the linearity range for
under identical conditions. The chromatograms showed a general observing PAHs is equivalent to 20-200 ng. The linearity of the
progression from high mass-high response to low mass-low PAHs is seen to have similar gradients illustrating comparable
response, and all peaks were present in the relative amounts in response-to-weight ratios and therefore the PAHs are directly
all chromatograms. Linearity was confirmed in the plot of total comparable at all weights of urban dust along the linearity. The
chromatographic area, relative to a blank run under identical high molecular weight PAHs benzo[ghi]perylene and indeno[1,2,3-
Analytical Chemistry, Vol. 72, No. 15, August 1, 2000 3565
Figure 4. Chromatographic linearity of 14 n-alkanes from C20-C33: C20 R2 ) 0.980, C21 R2 ) 0.931, C22 R2 ) 0.972, C23 R2 ) 0.984, C24 R2
) 0.981, C25 R2 ) 0.972, C26 R2 ) 0.984, C27 R2 ) 0.993, C28 R2 ) 0.987, C29 R2 ) 0.992, C30 R2 ) 0.972, C31 R2 ) 0.991, C32 R2 ) 0.976,
and C33 R2 ) 0.978.

Figure 5. Chromatographic linearity of 10 PAHs (phenanthrene R2 ) 0.983, pyrene R2 ) 0.982, fluoranthene R2 ) 0.973, benz[a]anthracene
R2 ) 0.996, chrysene R2 ) 0.997, benzo[b]fluoranthene R2 ) 0.992, benzo[e]pyrene R2 ) 0.995, benzo[a]pyrene R2 ) 0.976, benzo[ghi]-
perylene R2 ) 0.983, indeno[1,2,3-cd]pyrene R2 ) 0.971) in the Standard Reference Material SRM1649a Urban Dust.

cd]pyrene have the largest gradient discrepancy from the other
PAHs, indicating that comparisons of mass/response ratios
between these two PAHs and the lower molecular weight PAHs
will vary depending on the mass of the sample.
In order that the PAHs measured in the urban dust SRM can
be quantified, an internal deuterated standard, phenanthrene-d10
for the lower volatility certified PAHs phenanthrene, pyrene,
fluoranthene, benz[a]anthracene and chrysene, was utilized. The
higher molecular weight PAHs were also quantified relative to
3566 Analytical Chemistry, Vol. 72, No. 15, August 1, 2000
the phenanthrene-d10 standard by utilizing a NIST PAH standard
in DCM run with a phenanthrene-d10 standard, and response
factors were calculated. Prior to spiking the urban dust SRM
samples with phenanthrene-d10, a series of 1-µL samples of
phenanthrene-d10 of differing concentrations in DCM were ther-
mally desorbed from blank liners. A plot of chromatographic area
of phenanthrene-d10 against phenanthrene-d10 sample weight is
shown in Figure 6. Good linearity is again seen in the phenan-
threne-d10 series (R2 ) 0.963) over a range of weights comparable
Table 1. Comparison of Calculated Concentrations (mg kg-1) of Eight PAHs over Five Runs, Each Using
Approximately 3 mg of Urban Dust

( (confidence certified ( (confidence

mg kg-1 run 1 run 2 run 3 run 4 run 5 mean SD (σ) 95%) value 95%)
phenanthrene 4.658 4.502 4.815 4.657 4.472 4.621 0.124 0.109 4.14 0.37
fluoranthene 6.378 6.828 6.688 6.576 6.868 6.668 0.178 0.156 6.45 0.18
pyrene 5.438 4.898 5.008 4.938 4.945 5.055 0.199 0.175 5.29 0.25
benz[a]anthracene 2.523 2.238 2.142 3.478 2.956 2.859 0.191 0.167 2.21 0.073
chrysene 3.455 3.318 3.925 3.365 3.700 3.655 0.228 0.200 3.049 0.06
benzo[b]fluoranthene 6.030 5.410 5.656 7.372 6.536 6.201 0.780 0.684 6.450 0.640
benzo[e]pyrene 3.913 3.322 3.992 2.671 4.061 3.592 0.652 0.571 3.090 0.190
benzo[a]pyrene 2.830 2.103 2.480 2.392 4.141 2.789 0.922 0.809 2.509 0.087

the range of sample weights (1-5 mg) as well as for a series of

µ PAHs with 95% confidence limits coincide with the certified ranges
Figure 6. Chromatographic linearity (R2 ) 0.963) of phenanthrene-
d10 in DCM thermally desorbed after injection into a blank liner.
to that of the PAHs in the urban dust (0.05-0.25 µg; Figure 6).
Selection of a weight of phenanthrene-d10 within this linearity could
now be employed as an internal standard for the quantification of
the PAHs.
Quantification of PAHs in the Urban Dust SRM. Five
samples of ∼3 mg of urban dust were prepared and spiked with
1 µL of 150 ppm (0.15 µg) samples of phenanthrene-d10 in DCM
and then thermally desorbed in the usual way. The PAH peaks
were then integrated and quantified relative to the phenanthrene-
d10 peak. The quantified values are shown along with the certified
values in Table 1. The quantified values are extremely good, and
all of the calculated ranges (95% confidence) overlap with the
certified ranges except for chrysene. Calculated 95% confidence
limits are largest the greater the ∆Rt () Rt(target) - Rt(phenanthrene-d10))
as system responses vary across the chromatograms. Applying a
response factor adjusts for decrease in response between the
standard and the target PAH, but these higher PAHs also suffered
from variable molecular ion percentages over five quantification
runs. This is due to low concentrations being measured, the
reduction in response toward the high end of the chromatogram
as well as increased background interference from the UCM of
the chromatogram. Therefore, the PAHs of MW 276 have not been
quantified.
CONCLUSIONS
We have shown thermal desorption GC/MS to be a technique
that can accurately quantify polycyclic aromatic hydrocarbons in
an airborne particulate standard reference material using small
amounts (1-5 mg) of material.
We have shown the following: (1) very good linearity for the
total chromatographic response of the SRM (R2 ) 0.948) over
n-alkanes (C20-C33, R2 g 0.931) in the microgram range and 10
PAHs (MW ) 178-276, R2 g 0.971) in the nanogram range. The
linearity plots of the PAHs showed similar gradients illustrating
comparability of all PAHs relative to a standard at all sample
weights within the range. The higher PAHs could be quantified
by applying response factors relative to the phenanthrene-d10
standard; (2) a rough calculation of the total desorbable organic
fraction (3.7%) is comparable to the extractable (Soxhlet) fraction
certified as (4.6 ( 0.4%); (3) calculated concentrations of the eight
apart from chrysene, which falls close to, but just outside of, the
narrow certified range; (4) highly reproducible PAH concentra-
tions (MW ) 178-228, SD e0.228 mg kg-1, e7%; MW ) 252 SD
e0.922 mg kg-1, e33%) over five separate determinations using
sample weights in the middle of the linearity range.
A simple calculation using the average deposit weights col-
lected using low-volume sampling equipment in the London Air
Quality Network (LAQN) and total organic carbon (TOC) calcula-
tions, carried out in conjunction with LECO Instruments (UK)
Ltd., show that short sampling periods of 24-48 h provide
sufficient material to fall within the linearity of the technique
(range from 30 to 60% TOCsSRM calculated as 13-16% (certified
as 16 ( 5%)).
The technique is at present being used for the routine
determination of PAHs and other organic compounds in PM10 at
sites in London, providing much needed information on spatial
and temporal trends in organic composition of airborne particulate
matter, insight into potentially toxic compounds, and estimates
of the major contributing factors.
ACKNOWLEDGMENT
The authors thank the NERC URGENT project for the funding
to do this work and also South East Institute of Public Health
(SEIPH), the City of Westminster Council, the London Borough
of Greenwich Council, Pembrokeshire County Council, and
Dianne Hall (BP) for their continued support of this work. We
also thank LECO Instruments (UK) for their help in the TOC
analysis work.
Received for review November 29, 1999. Accepted May
17, 2000.
AC991372X

Analytical Chemistry, Vol. 72, No. 15, August 1, 2000 3567